CMSC 330: Organization of Programming Languages

1. Functional Programming with OCaml CMSC 330: Organization of Programming Languages

2. CMSC 330 Review Recursion is how all looping is done OCaml can easily pass and return functions

3. CMSC 330 The Call Stack in C/Java/etc. void f(void) { int x; x = g(3); } x <junk> 4 f x 3 g y 4 <junk> int g(int x) { int y; y = h(x); return y; } z 3 h int h (int z) { return z + 1; } int main(){ f(); return 0; }

4. CMSC 330 Nested Functions In OCaml, you can define functions anywhere Even inside of other functions let sum l = fold ((fun (a, x) -> a + x), 0, l)‏ let pick_one n = if n > 0 then (fun x -> x + 1)‏ else (fun x -> x - 1)‏ (pick_one -5) 6 (* returns 5 *)‏

5. CMSC 330 Nested Functions (cont’d)‏ You can also use let to define functions inside of other functions let sum l = let add (a, x) = a + x in fold (add, 0, l)‏ let pick_one n = let add_one x = x + 1 in let sub_one x = x - 1 in if n > 0 then add_one else sub_one

6. CMSC 330 How About This? (Equivalent to...)‏ takes a number n and list l and adds n to every element in l let addN (n, l) = let add x = n + x in map (add, l)‏ Accessing variable from outer scope let addN (n, l) = map ((fun x -> n + x), l)‏

7. CMSC 330 Consider the Call Stack Again How does add know the value of n? Read it off the stack? Possible, but the semantics are wrong (see above)‏ OCaml uses static scoping like C, C++, Java, and Ruby let map (f, n) = match n with [] -> [] | (h::t) -> (f h)::(map (f, t))‏ n 3 l <list> let addN (n, l) = map (add, l)‏ f <add> let add x = n + x in n x 1 addN (3, [1; 2; 3])‏

8. CMSC 330 Static Scoping In static or lexical scoping, (nonlocal) names refer to their nearest binding in the program text Going from inner to outer scope C example: In our example, add accesses addN’s n Refers to the x at file scope – that’s the nearest x going from inner scope to outer scope in the source code int x; void f() { x = 3; } void g() { char *x = "hello"; f(); }

9. CMSC 330 Returned Functions As we saw, in OCaml a function can return another function as a result So consider the following example When the anonymous function is called, n isn’t even on the stack any more! We need some way to keep n around after addN returns let addN n = (fun x -> x + n)‏ (addN 3) 4 (* returns 7 *)‏

10. CMSC 330 Environments and Closures An environment is a mapping from variable names to values Just like a stack frame A closure is a pair (f, e) consisting of function code f and an environment e When you invoke a closure, f is evaluated using e to look up variable bindings

11. CMSC 330 Example let add x = (fun y -> x + y)‏ → 3 + 4 (add 3) 4 → <closure> 4 → 7

12. CMSC 330 Another Example let mult_sum (x, y) = let z = x + y in fun w -> w * z (mult_sum (3, 4)) 5 → <closure> 5 → 5 * 7 → 35

13. CMSC 330 Yet Another Example let twice (n, y) = let f x = x + n in f (f y)‏ twice (3, 4)‏ → <closure> (<closure> 4)‏ → <closure> 7 → 10

14. CMSC 330 Still Another Example let add x = (fun y -> (fun z -> x + y + z))‏ (((add 1) 2) 3)‏ → ((<closure> 2) 3)‏ → (<closure> 3)‏ → 1+2+3

15. CMSC 330 Currying We just saw another way for a function to take multiple arguments The function consumes one argument at a time, creating closures until all the arguments are available This is called currying the function Discovered by Schönfinkel and Frege Named after the logician Haskell B. Curry Also called partial evaluation

16. CMSC 330 Curried Functions in OCaml OCaml has a really simple syntax for currying This is identical to all of the following: Thus: add has type int -> (int -> int)‏ add 3 has type int -> int The return of add x evaluated with x = 3 add 3 is a function that adds 3 to its argument (add 3) 4 = 7 This works for any number of arguments let add x y = x + y let add = (fun x -> (fun y -> x + y))‏ let add = (fun x y -> x + y)‏ let add x = (fun y -> x+y)‏

17. CMSC 330 Curried Functions in OCaml (cont’d)‏ Because currying is so common, OCaml uses the following conventions: -> associates to the right Thus int -> int -> int is the same as int -> (int -> int)‏ function application associates to the left Thus add 3 4 is the same as (add 3) 4

18. CMSC 330 Another Example of Currying A curried add function with three arguments: The same as Then... add_th has type int -> (int -> (int -> int))‏ add_th 4 has type int -> (int -> int)‏ add_th 4 5 has type int -> int add_th 4 5 6 is 15 let add_th x y z = x + y + z let add_th x = (fun y -> (fun z -> x+y+z))‏

19. CMSC 330 Currying and the map Function Examples let negate x = -x map negate [1; 2; 3] (* returns [-1; -2; -3 ] *)‏ let negate_list = map negate negate_list [-1; -2; -3] let sum_pairs_list = map (fun (a, b) -> a + b)‏ sum_pairs_list [(1, 2); (3, 4)] (* [3; 7] *)‏ What's the type of this form of map? let rec map f l = match l with [] -> [] | (h::t) -> (f h)::(map f t)‏ map : ('a -> 'b) -> 'a list -> 'b list

20. CMSC 330 Currying and the fold Function let add x y = x + y fold add 0 [1; 2; 3] let sum = fold add 0 sum [1; 2; 3] let next n _ = n + 1 let length = fold next 0 (* warning: not polymorphic *)‏ length [4; 5; 6; 7] What's the type of this form of fold? let rec fold f a l = match l with [] -> a | (h::t) -> fold f (f a h) t fold : ('a -> 'b -> 'a) -> 'a -> 'b list -> 'a

21. CMSC 330 Another Convention New keyword: function can often be used instead of match function declares an anonymous function of one argument Instead of It could be written let rec sum l = match l with [] -> 0 | (h::t) -> h + (sum t)‏ let rec sum = function [] -> 0 | (h::t) -> h + (sum t)‏

22. CMSC 330 Another Convention (cont’d)‏ Instead of It could be written let rec map f l = match l with [] -> [] | (h::t) -> (f h)::(map f t)‏ let rec map f = function [] -> [] | (h::t) -> (f h)::(map f t)‏

23. CMSC 330 Currying is Standard in OCaml Pretty much all functions are curried Like the standard library map, fold, etc. OCaml plays a lot of tricks to avoid creating closures and to avoid allocating on the heap It's unnecessary much of the time, since functions are usually called with all arguments How does this compare to C or Ruby?

24. CMSC 330 Higher-Order Functions in C C has function pointers but no closures (gcc has closures)‏ typedef int (*int_func)(int); void app(int_func f, int *a, int n) { int i; for (i = 0; i < n; i++)‏ a[i] = f(a[i]); } int add_one(int x) { return x + 1; } int main() { int a[] = {1, 2, 3, 4}; app(add_one, a, 4); }

25. CMSC 330 Higher-Order Functions in Ruby Use yield within a method to call a code block argument def my_collect(a)‏ b = Array.new(a.length)‏ i = 0 while i < a.length b[i] = yield(a[i])‏ i = i + 1 end return b end b = my_collect([1, 2, 3, 4, 5]) { |x| -x }

26. CMSC 330 Higher-Order Functions in Java/C++ An object in Java or C++ is kind of like a closure it’s some data (like an environment)‏ along with some methods (i.e., function code)‏ So objects can be used to simulate closures

27. CMSC 330 Exercises let times x y = x * y in let mystery w = times w in mystery 42 2 What is the type of times? What is the type of mystery? What is the final return value?

28. CMSC 330 Exercises let times x y = x * y in let mystery w = times w in mystery 42 2 What is the type of times? int -> int -> int What is the type of mystery? int -> int -> int What is the final return value? 84

29. CMSC 330 Exercises let bigger x y = if y > x then y else x in let biggest = function h::t -> fold bigger h t in biggest [1; 2; 3; 2; 1] What is the type of bigger? What is the type of biggest? What is the final return value?

30. CMSC 330 Exercises let bigger x y = if y > x then y else x in let biggest = function h::t -> fold bigger h t in biggest [1; 2; 3; 2; 1] What is the type of bigger? 'a -> 'a -> 'a What is the type of biggest? 'a list -> 'a What is the final return value? 3

31. CMSC 330 Exercises let foo a b = (a, b) in let bar x = map (foo x) in bar 4 [1; 2; 3] What is the type of foo? What is the type of bar? What is the final return value?

32. CMSC 330 Exercises let foo a b = (a, b) in let bar x = map (foo x) in bar 4 [1; 2; 3] What is the type of foo? 'a -> 'b -> 'a * 'b What is the type of bar? 'a -> 'b list -> ('a * 'b) list What is the final return value? [(4, 1); (4, 2); (4, 3)]